01
Definition
Automation is the use of technology to perform processes with limited direct human intervention. A complete automated system can sense conditions, estimate state, execute logic, control actuators, communicate with other systems, and report its status.
Control engineering is the discipline concerned with making dynamic systems behave according to desired objectives. Automation is broader: it includes control but also sequencing, supervision, instrumentation, communication, safety, diagnostics, and human-machine interaction.
environment → sensors → estimation/logic → controller → actuators → environment
02
Feedback
Feedback compares measured system behavior with a desired reference and uses the resulting error to determine corrective action.
e(t) = r(t) - y(t)
Negative feedback can reject disturbances, reduce sensitivity to model uncertainty, and improve tracking. Poorly designed feedback can instead create oscillation or instability.
03
The Automation Stack
- Field layer: sensors, transmitters, valves, motors, relays, drives, and actuators.
- Control layer: PLCs, PACs, embedded controllers, motion controllers, and regulatory control loops.
- Supervisory layer: SCADA, HMIs, alarms, historian systems, and operator interfaces.
- Plant layer: manufacturing execution, production planning, quality, maintenance, and asset management.
- Enterprise layer: analytics, planning, business systems, and cloud services.
04
Industrial Automation
Industrial automation applies computation and control to manufacturing, energy, water, transportation, logistics, buildings, laboratories, and process industries.
- Discrete automation handles events and individual manufactured units.
- Process automation regulates continuously varying physical quantities.
- Batch automation combines recipe-driven sequences with process control.
- Motion automation coordinates motors and mechanical systems.
05
Intelligent Automation
Modern automation increasingly incorporates machine learning, computer vision, optimization, predictive maintenance, anomaly detection, natural-language interfaces, and autonomous decision support.
The safest architecture often separates probabilistic or learned components from deterministic control and safety functions. A machine-learning model can recommend an action while a conventional controller and safety layer enforce physical constraints.